Reviewer's report 1
Han Liang, Department of Bioinformatics & Comp Biology, University of Texas M. D. Anderson Cancer Center (nominated by Laura Landweber)
Understanding the evolutionary patterns in the context of biological pathways is a topic of great importance. Wu and colleagues performed a comprehensive analysis on a well-characterized pathway in yeasts, HOG-signaling pathway. They found that the evolutionary rate of different components shows great variation, and the variation is influenced by many factors. This is an interesting study, providing useful insights into the evolution of a key pathway.
My comments are as follows:
1. As for the distribution of HOG pathway genes in the yeast phylogeny, does the gene duplicability (i.e., gene copy gain or loss) correlate with the pathway position?
Author's response: There are no correlation was observed between pathway position and gene duplicability. We observed the gene duplication events in HOG pathway mostly occurred in post-WGD species.
2. The authors detected positive selection on MSB2 gene. Is it possible to pinpoint the amino acids under adaptation and infer the functional effect of the amino acid changes?
Author's response: Thanks for the valuable suggestion. Besides the fast evolution of MSB2 gene, we also find that there is large insert/deletion in some strains of Saccharomyces cerevisiae. We will investigate the functional effect of these sequence changes in future work.
3. The authors reported that the MAPK genes are under stronger functional constraints and that the evolutionary rate is negatively correlated with the positions in the pathway (positions 1-6). These two points are actually redundant because the MAPK (regulator module) is in the downstream of sensor module. So within a module, does the evolutionary rate depend on the position?
Author's response: According to an engineering view of dynamical control, gene function in a signaling pathway can be classified into three functional roles: "sensor", "regulator", "actuator". In each module, the downstream elements have higher levels of selective constraint than the upstream elements. For regulator module, ρ = -0.89, P = 0.006 (Spearman's rank correlation). The correlation of other two modules is more clearly (Figure 3).
4. In the last section, the authors examined the correlations of dN/dS with a series of factors. Since these factors are often interrelated, it would be better to use principal component analysis or partial correlation analysis to estimate the relative contributions of each factor?
Author's response: We agree with the reviewer that the suggested statistical analysis should be better to estimate the relative contributions of each factor. In this study, we used simple Spearman rank correlation analysis to estimate the correlation, and the path analysis to estimate the indirect effects of these factors. Therefore, we tried to findout which factors affect the evolution, but not considering which factor play main role in the evolution.
Minor points:
"Most genes have orthologs in all the yeast species and were identified as unique singletons in most cases." How can 8 out of 19 genes become "most"?
Author's response: There are 10 yeast species surveyed in this study. The total number of genes is 190. As shown in Table 1, the singleton gene number is 169 (190 - (15 + 2 + 4)). So we say that most genes were singleton.
I declare that I have no competing interests.
Reviewer's report 2
Georgy Bazykin, Department of Bioinformatics, Russian Academy of Science (nominated by Mikhail Gelfand)
The manuscript by Wu et al. is an addition to the existing literature on correlations between the pathway position of the gene and its evolutionary rate. The authors show that the genes located downstream in HOG pathway are under stronger negative selection.
My main concern is that there seems to be a number of flaws in the statistical analyses supporting the authors' main conclusions.
1. Multiple testing. The authors test all 19 genes of the HOG pathway with the free-ratio model of codeml (apparently for a total of ~19*7 comparisons), and identify a single gene at a single branch (MSN2 of Calb) with dN/dS > 1. They then test MSB2 gene for positive selection using LRT, and detect positive selection. However, due to the sheer number of multiple tests involved, detecting a single gene with dN/dS > 1 could occur due to chance effects. There is no built-in correction for multiple testing in codeml (Yang 1998 Mol Biol Evol 15(5):568-573), and the results of multiple comparisons should be interpreted with more caution. In Methods, the authors write that they used the FDR rate of 0.05, but this apparently only applies to comparisons of codons using M7 and M8 models (which apparently is never implemented, see below).
Author's response: We agree with the reviewer, and as this comment says, we first test all 19 genes of the pathway with the free-ratio model, then identify positive selection using likelihood ratio test (LRT) test to compare M7 and M8 model. However, the gene with d
N
/d
S
>1 is not used as evidence to detect positive selection, which is applied by the comparisons of codons using M7 and M8 models in this study. We have described this analysis in Methods (Estimation of evolutionary rate and detection of positive selection). Furthermore, considering 1 out of 19 tests to be positive, and FDR cutoff = 0.05, the P value should be P < 0.0026 (1 * 0.05/19). As the MSB2 with P < 0.001, so it's selected as positive selection gene.
2. Arbitrariness in selection of data. In analysis of the correlation between the pathway position and dN/dS, the authors exclude genes in position 7 and 8 of HOG pathway. Why these genes should be excepted is not explained, and the authors include them in the rest of the tests.
Author's response: The exception of position 7 & 8 could be explained by two reasons: first, it made sense that the external signal is transmitted into nuclear from Sln1/Sho1 to Hog1, which changed gene transcription. The HOG signal transduction process is restricted in genes from SLN1 to HOG1, but transcription factors are involved in many other biological processes. Second, the increased divergence of transcription factors in postion 7 had been observed. If the position 7 and 8 was included, the correlation didn't hold again.
I declare that I have no competing interests.
Reviewer's report 3
Zhenguo Lin, Department of Ecology and Evolution, The University of Chicago (nominated by John Logsdon)
Different organisms may respond differently to the specific stress. It is of great interest to study how such difference was evolved by examining the evolution of the related stress response pathways. The yeast S. cerevisiae cells respond to increased intracellular osmolarity by activating of the high osmolarity glycerol (HOG) MAPK pathway. This manuscript indentify the orthologous genes of HOG pathway from 10 closely related yeast species, and examined the evolutionary rates of genes at each step of this pathway. The authors also investigated the other factors that may affect the evolutionary rates of these genes. The authors found that the HOG pathway is conserved in hemiascomycete yeasts, but the evolutionary rates vary greatly in different lineages. One interesting observation is that the evolutionary rates of genes tend to be negatively correlated with their positions in the pathway, suggesting a stronger selection constrains at the downstream genes. Overall, this is a well-executed analysis and well organized manuscript, although more work need to be done to improve the quality of writing.
Major comments
1. The authors claimed that the MSB2 genes have been under positive selection. Although the ω = 0.2496 value is probably the largest among all genes in the pathway, it is still far below 1. However, if MSB2 was under positive selection does not affect the conclusion of this paper.
Author's response: Although the d
N
/d
S
value of the whole sequence of MSB2 is below 1, but some sites (codons) of this gene may evolve fast and under positive selection. We used the site model M7 and M8 comparison to detect positively selected genes, which allow d
N
/d
S
to vary among codons.
2. In the introduction, the authors said C. albicans and A. gossypii have different sensitivity responding to environmental osmolarity stress. Are the evolutionary rates of HOG genes different between the two species? It would make this study more valuable if the results are correlated with different osmolarity stress responses among different species.
Author's response: Thanks for the reviewer's useful comments. We did observe the difference of evolutionary rates of these two species (Wilcoxon rank sum test, P = 0.00213). However, the omostic stress resistance have been compared in only a few species, and the resistance capability cannot be quantified, so we cannot analysis the correlation relationship between evolutionary rate and stress resistance quantitatively.
3. The Figure 5 shows a negative correlation between the dN/dS value and the position during the HOG signal transduction process. As said by authors, the evolutionary rates are quite different among different lineages. I am not sure if it is a good idea to mix data from all lineages to study the correlation. Does the correlation still exist in each individual species or how many species have such pattern?
Author's response: Following the reviewer's suggestion, we have considered evolutionary rate of each lineage of the surveyed yeast taxa, and found the pattern of d
N
/d
S
variation correlated significantly with pathway position still holds in S. parodoxus, S. miketae and A. gossypii.
Minor comments
" are characterized extensively" → "have been characterized extensively"
Author's response:Following the reviewer's suggestion, we haved changed " are characterized extensively" to "have been characterized extensively".
"10 yeast genomes" → "10 yeast species"
Author's response: Following the reviewer's suggestion, we haved changed " 10 yeast genomes " to "10 yeast species ".
"Among them" is not clear to the context. "them" could be misleading as the three functional modules.
Author's response: Following the reviewer's suggestion, we have removed these two words.
fungi is more "evolutionary conserved" is that true? any reference?
Author's response: Following the reviewer's suggestion, we have added two references."proofed" proof is not a verb
Author's response: Following the reviewer's suggestion, we have changed " proofed " to " proved ".
"The aligned amino acid sequences were reversely translated into nucleotide sequences by in-house Perl script for further analysis". Did you mean convert protein alignment into DNA alignment using their original DNA sequence? It is impossible to "Translate" amino acid sequences to translated nucleotide.
Author's response: Following the reviewer's suggestion, we have changed " translated " to " converted ".
" we generated 19 HOG-MAPK pathway genes" it is not appropriate to use "generated" here.
Author's response: We agree with the reviewer and have changed "generate" to "identified".
" their surrounding genes" it is misleading. it could be neighboring genes on the chromosome.
Author's response: Following the reviewer's suggestion, we have add "in HOG pathway" in this sentence.
" Several earlier studies", only one reference is listed.
Author's response: Following the reviewer's suggestion, we have added two references.
"fungi" → fungal
Author's response: Following the reviewer's suggestion, we have changed "fungi " to " fungal ".
I declare that I have no competing interests.